A Method, System and Medium for Construction Information Management in Mechatronic Engineering

Abstract

The present application discloses a method, system and medium for construction information management of mechanical and electrical engineering, which relates to the field of mechanical and electrical engineering and includes: collecting an image set of the construction site of mechanical and electrical engineering; performing feature recognition on the image set; dividing the risk levels of construction personnel's behaviors and annotating material attributes according to the recognition results; dividing the construction process into multiple stages, constructing a Bayesian network model of time slices for each stage, adding a temporal dependence relationship between the Bayesian network models of adjacent time slices, and establishing a dynamic Bayesian network risk evolution model for different stages of the construction process; using the dynamic Bayesian network risk evolution model to predict the risk distribution heat map of each stage of construction in a future period of time; and performing construction management according to the risk distribution heat map. Aiming at the static risk assessment using traditional Bayesian inference in the prior art, the present application improves the accuracy and real-time performance of the construction risk prediction of mechanical and electrical engineering.

Description

Technical Field

[0001] This application relates to the field of mechatronic engineering, and particularly to a method, system, and medium for managing construction information in mechatronic engineering. Background Art

[0002] With the rapid development of modern technology and the increasing complexity of engineering projects, the construction of mechatronic engineering faces more and more risks and uncertain factors. Timely and accurately identifying and predicting potential risks during the construction process is of great significance for ensuring project quality, controlling costs and schedules, and ensuring construction safety.

[0003] Traditional risk management methods, such as fault tree analysis and event tree analysis, mainly use static risk assessment models, set model parameters through expert experience, and conduct qualitative or semi - quantitative analysis of construction risks. These methods are difficult to capture and predict the dynamic changes of risk factors during the construction process, and the subjectivity and uncertainty of the assessment results are relatively large, unable to meet the growing demand for risk management in mechatronic engineering construction.

[0004] In recent years, Bayesian networks have been widely used in the field of risk assessment due to their advantages in uncertainty reasoning and causal relationship modeling. By constructing a Bayesian network containing risk factor nodes and obtaining conditional probability parameters through expert knowledge or data learning methods, the probability of risk events occurring can be inferred and predicted. However, most traditional Bayesian networks are static models, assuming that risk factors are independent of each other, and it is difficult to depict complex causal relationships and time dynamics, resulting in less than ideal accuracy in risk assessment and prediction. Summary of the Invention

[0005] Aiming at the low assessment accuracy caused by using traditional Bayesian inference for static risk assessment in the prior art, this application provides a method, system, and medium for managing construction information in mechatronic engineering. By constructing a dynamic risk assessment and prediction model suitable for the mechatronic engineering construction process, the accuracy and real - time performance of mechatronic engineering construction risk prediction are improved.

[0006] The objectives of this application are achieved through the following technical solutions.

[0007] One aspect of the present application provides a method for managing construction information in mechanical and electrical engineering, including: S1, collecting an image set of the mechanical and electrical engineering construction site; S2, performing feature recognition on the image set, including recognizing the behaviors and materials of construction workers, to obtain a recognition result; S3, according to the recognition result, dividing the risk levels of construction worker behaviors and labeling the material attributes; S4, dividing the construction process into multiple stages, constructing a Bayesian network model of time slices for each stage, adding a temporal dependence relationship between the Bayesian network models of adjacent time slices, and establishing a dynamic Bayesian network risk evolution model for different stages of the construction process; wherein, taking the risk levels of construction worker behaviors and material attributes as Bayesian network nodes, setting the conditional probability distribution between the nodes, and calculating the risk probability distribution under various factor combinations through the Bayesian network inference algorithm; S5, using the dynamic Bayesian network risk evolution model to predict the risk distribution heat map of each stage of the construction in the future for a period of time; S6, performing construction management according to the risk distribution heat map. Traditional risk assessment methods usually only give an overall risk level, lacking an explanation of the risk causes and distributions. By generating a risk distribution heat map, the present application intuitively shows the risk level distributions at different time stages and different regions, and can trace the state changes of risk nodes, enabling managers to quickly identify risk hotspots and formulate targeted control measures according to the risk causes, improving the interpretability and operability of the risk assessment results.

[0008] Further, in S2, performing feature recognition on the image set, including recognizing the behaviors and materials of construction workers, to obtain a recognition result, including: S21, using a convolutional neural network to extract the deep features of the image as the first feature; S22, extracting the texture features, edge features, and color features of the image as the second feature; only extracting the texture, edge, and color features of the image may be difficult to comprehensively depict the semantic information of the behaviors and materials of construction workers; S23, fusing the first feature and the second feature to obtain a fused feature vector; S24, using the fused feature vector as the input, performing forward propagation calculation using a pre-trained convolutional neural network, and judging the types of construction worker behaviors and material types in the image according to the class probabilities of the output layer.

[0009] On the one hand, the images at the mechanical and electrical construction site usually contain rich semantic information, such as the behavior postures of construction workers, the types and states of materials, etc. Only extracting the low-level features of the image (such as texture, edge, and color) may be difficult to fully express these high-level semantic contents. Through multi-level feature extraction and abstraction, the convolutional neural network can automatically learn the deep features in the image, capturing the key information and internal patterns of the construction scene. The fusion of deep features and low-level features can more comprehensively depict the semantic features of mechanical and electrical construction images.

[0010] On the other hand, the electromechanical construction environment is complex and variable, and the behaviors of construction workers and the states of materials show diversity and variability. Traditional feature extraction methods may be difficult to handle this complexity, resulting in low recognition accuracy. Through end-to-end training, convolutional neural networks can adaptively learn and optimize feature representations, capturing key patterns and discriminative information in construction scenarios. The introduction of deep features can significantly improve the accuracy of behavior and material recognition, especially for small target detection and fine-grained recognition in complex backgrounds.

[0011] Furthermore, S3, according to the recognition results, classify the behavior risk levels of construction workers and label the material attributes, including: classifying the behavior risks into three levels: low risk, medium risk, and high risk according to the behavior types of construction workers, where the behavior risks include not wearing safety equipment and non-standard operations; labeling the types and quantities of materials according to the material types.

[0012] This application classifies the behavior risks of construction workers into three levels: low risk, medium risk, and high risk, and labels the material attributes as types and quantities. Essentially, it transforms continuous behavior and material characteristics into discrete variable representations. On the one hand, Bayesian networks are a type of probabilistic graphical model used to describe the conditional dependencies and joint probability distributions between variables. In a Bayesian network, each node represents a random variable, and the edges represent the causal relationships or correlations between variables. Bayesian networks usually assume that the variables are discrete, and each variable has a finite number of possible states or values. Transforming continuous behavior risks and material attributes into discrete levels and categories can meet the modeling requirements of Bayesian networks, facilitating subsequent network construction and parameter learning.

[0013] On the other hand, if continuous behavior and material characteristics are directly used as nodes in a Bayesian network, continuous probability distributions such as Gaussian distributions and exponential distributions need to be processed. In this case, the computational complexity of the inference algorithm is relatively high, making it difficult to achieve real-time risk assessment and prediction. Transforming continuous features into discrete variables can significantly reduce the computational complexity of Bayesian inference. The conditional probability distribution of discrete variables can be represented by a finite-dimensional probability table or factor, and the inference process can be completed through table lookup and simple arithmetic operations, with a low time complexity. Discretization can also reduce the sensitivity to data quality and noise, improving the stability and robustness of the inference results.

[0014] Finally, in the field of mechanical and electrical construction, the assessment of personnel behavior risks and material properties usually relies on domain knowledge and expert experience. Dividing continuous features into discrete levels and categories can better utilize the prior knowledge and experience rules of domain experts. For example, experts can summarize the risk level classification criteria under different behavior patterns based on experience, such as "not wearing a safety helmet" corresponding to high risk, "not fastening the seat belt" corresponding to medium risk, etc. Similarly, the classification of material types and quantities can also be based on professional standards and practices. Introducing expert knowledge into the construction process of the Bayesian network can improve the accuracy and interpretability of risk assessment, and also facilitate the initialization and constraint of network parameters.

[0015] Further, in S4, the construction process is divided into multiple stages, and a Bayesian network model of time slices is constructed for each stage. Temporal dependence relationships are added between the Bayesian network models of adjacent time slices to establish a dynamic Bayesian network risk evolution model for different stages of the construction process, including: S41, according to the construction plan and progress, the entire construction process is divided into multiple consecutive time slices; each time slice represents a construction stage; S42, within each time slice, the personnel behavior risks during construction are divided into three levels: low risk, medium risk, and high risk, serving as the behavior risk nodes of the Bayesian network; the types and quantities of materials are marked according to the material types, serving as the material property nodes of the Bayesian network; a Bayesian network is constructed based on the behavior risk nodes and material property nodes; S43, temporal dependence edges are added between the Bayesian network of adjacent time slices to obtain a dynamic Bayesian network; among them, the temporal dependence edge represents the influence of the construction risk state at the previous moment on the current moment; S44, according to historical data, the conditional probability distribution parameters of the dynamic Bayesian network are learned to obtain a dynamic Bayesian network model.

[0016] Further, in S44, according to historical data, the conditional probability distribution parameters of the dynamic Bayesian network are learned to obtain a dynamic Bayesian network model, including: organizing and marking the historical data according to time slices, counting the state distributions of the behavior risk nodes and material property nodes in the historical data within each time slice to obtain the empirical probabilities of the node states; for the temporal dependence edges between adjacent time slices, counting the joint distribution of the node states at the previous moment and the current moment to obtain the empirical probabilities of state transitions; according to the state distributions and joint distributions, estimating the initial state transition probabilities between nodes in the dynamic Bayesian network as the initial values of the conditional probability distribution parameters; using the Expectation-Maximization (EM) algorithm and a random restart strategy to iteratively optimize the conditional probability distribution parameters of the dynamic Bayesian network; using the initial state transition probabilities to correct the iteratively optimized conditional probability distribution parameters to obtain the corrected conditional probability distribution parameters; using the final conditional probability distribution parameters to construct a dynamic Bayesian network model.

[0017] Furthermore, the Expectation-Maximization (EM) algorithm and the random restart strategy are adopted to iteratively optimize the conditional probability distribution parameters of the dynamic Bayesian network, including: setting the number of random restarts N, where a random restart means re-initializing the parameters during the iterative optimization process; for each random restart, randomly perturb the initial state transition probability to generate the initial parameter values of the EM algorithm; for each random restart, randomly perturb the initial state transition probability to generate the initial parameter values of the EM algorithm. This step introduces random perturbations to the initial parameters, making the starting point of each restart different, increasing the diversity of parameter search, helping to jump out of local optima, and finding better solutions. In the E step, using the current conditional probability distribution parameters, calculate the expected log-likelihood function of the observed data; the observed data includes the state distribution and state transitions; the observed data in this application includes the state distribution and state transitions, making full use of the temporal information in the dynamic Bayesian network. In the M step, maximize the expected log-likelihood function and update the estimates of the conditional probability distribution parameters; repeat the E step and the M step until the conditional probability distribution parameters converge or reach the maximum number of iterations; after completing N random restarts, select the result with the largest expected log-likelihood function value as the conditional probability distribution parameters after iterative optimization.

[0018] Traditional Expectation-Maximization (EM) algorithms have defects such as local optima and slow convergence speed in dealing with parameter estimation of dynamic Bayesian networks in mechatronics engineering, resulting in the conditional probability distribution parameters learned may not be the global optimal solution, affecting the accuracy of risk prediction. In this application, the number of random restarts N is set, representing the number of times of re-initializing the parameters during the iterative optimization process. This step introduces a random restart mechanism. By re-initializing the parameters multiple times and running the EM algorithm independently, the parameter search range is expanded, increasing the probability of finding the global optimal solution and overcoming the defect that traditional EM algorithms are prone to falling into local optima.

[0019] Furthermore, use the initial state transition probability to correct the conditional probability distribution parameters after iterative optimization to obtain the corrected conditional probability distribution parameters, including: taking the initial state transition probability as the parameter of the Dirichlet distribution to construct the prior distribution of the conditional probability distribution parameters; taking the conditional probability distribution parameters after iterative optimization as the observed data; according to Bayes' theorem, calculate the posterior distribution of the conditional probability distribution parameters based on the prior distribution and the observed data; use the mean or mode of the posterior distribution as the corrected conditional probability distribution parameters.

[0020] In the construction information management of mechatronic engineering, due to the complex and changeable construction environment and the uncertainty of risk factors, the conditional probability distribution parameters obtained only by data-driven parameter learning methods (such as the EM algorithm) may deviate from the actual situation, affecting the risk prediction performance of the dynamic Bayesian network model. In this application, by taking the initial state transition probability as the parameter of the Dirichlet distribution, the prior information of the conditional probability distribution parameters is introduced, making up for the deficiency of pure data-driven methods. The iteratively optimized conditional probability distribution parameters are used as the observed data. The conditional probability distribution parameters learned by making full use of the EM algorithm and the random restart strategy are used as the observed data, which reflects the risk evolution law mined from historical construction data. Using the basic principle of Bayesian inference, the prior distribution (expert experience) and the observed data (data-driven results) are fused. By calculating the posterior distribution of the conditional probability distribution parameters, a parameter estimation result that takes into account both subjective knowledge and objective data is obtained, realizing the optimization and correction of the initial parameters.

[0021] Further, in step S5, use the dynamic Bayesian network risk evolution model to predict the risk distribution heat map of each construction stage in the future for a period of time, including: S51, according to the construction plan and progress, determine the future time period to be predicted, and divide the future time period into continuous time slices, where each time slice represents a construction stage; S52, input the construction personnel behavior types and material types within each future time slice into the dynamic Bayesian network model as evidence nodes; S53, according to the states of the evidence nodes, use the conditional probability distribution parameters to calculate the posterior probability distributions of other risk nodes; S54, according to the posterior probability distributions of the risk nodes within each future time slice, generate a time-series probability distribution map of the construction risk level; S55, visualize the time-series probability distribution map of the construction risk level as a risk distribution heat map.

[0022] Further, in step S53, according to the states of the evidence nodes, use the conditional probability distribution parameters to calculate the posterior probability distributions of other risk nodes, including: taking the states of the evidence nodes as known conditions, and constructing a chain graph of the dynamic Bayesian network according to the corrected conditional probability distribution parameters; the chain graph consists of multiple time slices, and the nodes within each time slice represent construction risk factors and material attribute factors; starting from the end time slice of the chain graph, calculate the posterior probabilities of other risk nodes according to the states of the evidence nodes and the conditional probability distribution parameters; for the previous time slice, take the posterior probabilities of the risk nodes in the next time slice as the observed evidence, and combine the states of the evidence nodes and the conditional probability distribution parameters in the current time slice to calculate the posterior probabilities of other risk nodes in the current time slice; recursively calculate until reaching the starting time slice to obtain the posterior probability distributions of the risk nodes in all time slices.

[0023] This application adopts an interface tree algorithm to calculate the posterior probability distribution of other risk nodes based on the status of the evidence node. Different from the static method of assessing risk once, this method constructs a chain graph of a dynamic Bayesian network and uses the modified conditional probability distribution parameters to recursively calculate the posterior probability distribution of risk nodes in each time slice starting from the terminal time slice. It fully considers the causal relationship and transmission effect of risk in the time dimension, and obtains more accurate and comprehensive risk prediction results.

[0024] In the process of calculating the posterior probability of risk nodes, the status information of evidence nodes and the corrected conditional probability distribution parameters are fully utilized. Compared with the existing methods, when executing the interface tree algorithm, this method not only considers the status of evidence nodes in the current time slice, but also uses the posterior probability of risk nodes at the previous moment as observation evidence. Combined with the status of evidence nodes and conditional probability distribution parameters of the current time slice, the posterior probability of other risk nodes in the current time slice is recursively calculated, which fully utilizes the temporal dependency of the dynamic Bayesian network and improves the accuracy and reliability of risk prediction.

[0025] Another aspect of the present application also provides a mechanical and electrical engineering construction information management system for executing a mechanical and electrical engineering construction information management method of the present application.

[0026] Compared with the prior art, the advantages of this application are:

[0027] The traditional Bayesian network model has defects such as static description and independent and identically distributed assumptions in dealing with mechanical and electrical engineering construction management. It is difficult to characterize the dynamic evolution characteristics and temporal dependencies of risk factors in the construction process, resulting in low accuracy of risk assessment and prediction. In this application, the construction process is divided into multiple stages, and a Bayesian network model of a time slice is constructed for each stage. This breaks through the limitation that the traditional Bayesian network can only describe the state of the system at a single moment. By discretizing the construction process in the time dimension to form a series of time slices, a static Bayesian network is constructed in each time slice, realizing a dynamic description of the construction process. In each time slice, a Bayesian network is constructed according to the risk level of construction personnel behavior and material properties, comprehensively considering the two major risk sources of unsafe behavior of people and unsafe state of things. By dividing the behavioral risk into levels and marking the material properties, a Bayesian network containing behavioral risk nodes and material attribute nodes is constructed, realizing a comprehensive characterization of risk factors in a single time slice.

[0028] By adding time-dependent edges between the Bayesian network models of adjacent time slices, a dynamic Bayesian network is obtained. This overcomes the assumption of independent and identical distribution between time slices in the traditional Bayesian network. By introducing time-dependent edges that represent the impact of the construction risk status at the previous moment on the current moment, the causal relationship and transmission law of risks in the time dimension are characterized, thus realizing the modeling of the dynamic evolution of risks.

[0029] Learn the conditional probability distribution parameters of the dynamic Bayesian network based on historical data to obtain a complete dynamic Bayesian network model. Use the historical data accumulated in engineering construction to estimate the conditional probability distribution between nodes in the network through a parameter learning algorithm, breaking through the subjectivity of setting parameters in traditional Bayesian networks that rely on expert experience, making the model closer to reality and having better generalization ability. Brief Description of the Drawings

[0030] This application will be further described in the manner of exemplary embodiments, and these exemplary embodiments will be described in detail through the drawings. These embodiments are not restrictive. In these embodiments, the same numbers represent the same structures, where:

[0031] Figure 1 is an exemplary flowchart of a method for managing electromechanical engineering construction information shown in some embodiments of this application;

[0032] Figure 2 is an exemplary flowchart of dividing construction personnel behaviors and material types shown in some embodiments of this application;

[0033] Figure 3 is an exemplary flowchart of constructing a dynamic Bayesian network model shown in some embodiments of this application. Detailed Description of the Specific Embodiments

[0034] The following will describe in detail the methods and systems provided in the embodiments of this application with reference to the drawings.

[0035] As Figure 1 shown, collect the image set of the electromechanical engineering construction site; perform feature recognition on the image set, including recognizing the behaviors of construction personnel and materials, to obtain the recognition results; according to the recognition results, divide the risk levels of construction personnel behaviors and label the material attributes; divide the construction process into multiple stages, construct a Bayesian network model for each time slice in each stage, add temporal dependencies between the Bayesian network models of adjacent time slices, and establish a dynamic Bayesian network risk evolution model for different stages of the construction process; among them, use the risk levels of construction personnel behaviors and material attributes as the nodes of the Bayesian network, set the conditional probability distribution between the nodes, and calculate the risk probability distribution under various factor combinations through the Bayesian network inference algorithm; use the dynamic Bayesian network risk evolution model to predict the risk distribution heat map of each stage of construction in the future for a period of time.

[0036] S1. Deploy multiple high-definition digital cameras at the construction site of the mechanical and electrical engineering project to collect real-time images of the construction process in all directions and from multiple angles. The positions and quantities of the cameras are reasonably set according to the scale, layout, and key risk points of the construction site to ensure that the main areas and links of the construction can be covered. The collected images are automatically stored in a local or cloud-based image database at a certain frame rate (such as 10 frames per second) to form an image set of the construction site. At the same time, metadata information such as the acquisition time and position of each image is recorded to provide a spatio-temporal background for subsequent image recognition and risk analysis.

[0037] As Figure 2 shown, S2. Perform feature recognition on the image set, including recognizing construction worker behaviors and materials, and obtaining recognition results, including: S21. Select a pre-trained convolutional neural network model (such as VGG, ResNet, etc.), input the image into the network for forward propagation calculation. Extract the output of the intermediate layer of the network (such as the pooling layer before the fully connected layer) as the deep feature of the image, representing the high-level semantic information of the image. The deep feature is represented in the form of a high-dimensional vector and has strong discriminative ability and generalization ability.

[0038] S22. Perform traditional feature engineering on the image to extract manually designed low-level visual features. Use algorithms such as local binary pattern (LBP) and histogram of oriented gradients (HOG) to extract the texture features of the image, depicting the texture patterns and structural information of local regions of the image. Use operators such as Canny and Sobel to calculate the edge features of the image, highlighting the contour and boundary information of objects. Extract the color features of the image through methods such as color histogram and color moment to reflect the color distribution and dominant color information of the image.

[0039] S23. Integrate the deep features extracted in step S21 and the texture, edge, and color features extracted in step S22 to form a more comprehensive and robust feature representation. Adopt strategies such as feature concatenation and feature weighting to connect or linearly combine different types of feature vectors to generate a high-dimensional integrated feature vector. The integrated feature vector takes into account both high-level semantic information and low-level visual information and can better depict the content and attributes of the image.

[0040] S24, Select a convolutional neural network model pre-trained on a large-scale image dataset (such as ImageNet) as the infrastructure for image recognition, and perform fine-tuning and specialized customization for the construction scenario. Input the fused feature vectors generated in step S23 into the network, and obtain the class probability distribution of the output layer through forward propagation calculation. According to the level of class probability, determine whether there are construction workers in the image and the specific behavior types of the construction workers (such as normal operation, violation of regulations, lack of protection, etc.). At the same time, identify the main material types (such as steel bars, concrete, scaffolding, etc.) appearing in the image. Output the results of image recognition and store them in association with the metadata information of the image to form a structured recognition record.

[0041] S3, After completing the feature recognition of the construction site image in step S2, the recognition results including the behavior types and material types of the construction workers are obtained. According to these recognition information, further divide the behavior risk levels and label the material attributes, specifically as follows: Set the rules for dividing risk levels, and divide the behavior risks into three levels: low risk, medium risk, and high risk according to the behavior types of the construction workers.

[0042] Low-risk behaviors: Normal operation behaviors, such as wearing safety helmets properly, using tool equipment correctly, following operation procedures, etc., indicating that the behavior of construction workers is safe and compliant, and the risk level is low. Medium-risk behaviors: Failure to wear some safety equipment, such as not wearing safety ropes, safety shoes, etc., or the operation is not standardized enough, such as not fully following the operation steps, not using protective equipment correctly, etc., indicating that the behavior of construction workers has certain potential safety hazards, and the risk level is medium. High-risk behaviors: Serious violations of regulations, such as not using any safety protection equipment during high-altitude operations, smoking in dangerous areas, working under the influence of alcohol, etc., indicating that the behavior of construction workers is extremely likely to lead to safety accidents, and the risk level is high. According to the above division rules, determine the risk level of each identified construction worker behavior, generate behavior risk labels (such as "low risk", "medium risk", "high risk"), and store them in association with the original image and recognition record.

[0043] Label the attributes of the identified material types, mainly including the type and quantity information of the materials. Material type: Determine the main material types appearing in the image according to the recognition results, such as steel bars, concrete, wooden boards, sand and gravel, etc., and generate material type labels (such as "steel bars", "concrete", etc.). Material quantity: Combine image segmentation and object detection technologies to count different instances of the same material and estimate the quantity of the materials to generate quantity labels (such as "steel bars: 20 pieces", "concrete: 5 cubic meters", etc.). Store the material type and quantity labels in association with the original image and recognition record to form complete material attribute annotation information.

[0044] Such as Figure 3As shown in the figure, in S4, the construction process is divided into multiple stages. For each stage, a Bayesian network model of time slices is constructed, and temporal dependence relationships are added between the Bayesian network models of adjacent time slices to establish a dynamic Bayesian network risk evolution model for different stages of the construction process, including: S41. According to the construction plan and progress, the entire construction process is divided into multiple consecutive time slices, and each time slice represents a construction stage. Specifically, it can be divided according to the work breakdown structure (WBS) and critical path method (CPM) of the construction project. WBS decomposes the construction project into multiple manageable work packages, and CPM determines the logical relationships and time constraints between each work package. On this basis, the construction process is divided into several construction stages with continuous time and relatively independent content, and each stage corresponds to a time slice. The granularity of the time slice can be adjusted according to the requirements of construction management, such as being divided by day, week, month, etc.

[0045] S42. Within each time slice, a Bayesian network containing construction worker behavior risk nodes and material attribute nodes is constructed. According to the risk level division rules in step S3, the behavior risk nodes of construction workers are divided into three states: low risk, medium risk, and high risk. The low-risk state indicates that the behavior of construction workers complies with safety regulations and the risk level is relatively low; the medium-risk state indicates that there are certain unsafe factors in the behavior of construction workers and the risk level is medium; the high-risk state indicates that the behavior of construction workers seriously violates safety regulations and the risk level is high. Within each time slice, the state of the behavior risk node is determined according to the behavior data of construction workers in the corresponding stage and the risk level division rules.

[0046] According to the material management situation at the construction site, material attribute nodes are set to represent the attribute information of different materials. The state of the material attribute node can be determined according to the material type and quantity marking results. For example, for steel bar materials, two attribute nodes of "specification" and "quantity" can be set. The state of the specification node can be "qualified" or "unqualified", and the state of the quantity node can be "sufficient" or "insufficient". According to data such as material inspection upon arrival and inventory count, the specific state of the material attribute node within each time slice is determined.

[0047] Directed edges are added between the behavior risk nodes and the material attribute nodes to represent the causal dependence relationship between them. The direction of the edge is determined according to domain knowledge and data analysis results, usually pointing from the influencing factor to the affected object. For example, if the unqualified specification of steel bar materials may lead to an increase in the behavior risk of construction workers, then an edge is drawn from the "steel bar specification" node to the "behavior risk" node. The causal dependence relationship can be identified and verified through methods such as expert experience, accident case analysis, and statistical data mining.

[0048] Combine the behavioral risk nodes, material property nodes, and the causal dependency edges between them to form a complete Bayesian network structure. The Bayesian network represents the dependency relationships between nodes through a directed acyclic graph (DAG), where nodes represent random variables and edges represent conditional dependency relationships. Construct an independent Bayesian network within each time slice to reflect the causal relationship between the behavioral risks of construction workers and material properties during that stage.

[0049] Assign a conditional probability distribution (CPD) to each node in the Bayesian network, which represents the probability distribution of the node under different combinations of the states of its parent nodes. The conditional probability distribution can be initially determined through methods such as expert knowledge estimation and historical data statistics, and can be further optimized through parameter learning algorithms later (as described in step S44). For example, for the behavioral risk node, the following conditional probability distribution can be set: when the steel bar specifications are qualified and the quantity is sufficient, the low-risk probability is 0.8, the medium-risk probability is 0.15, and the high-risk probability is 0.05; when the steel bar specifications are unqualified or the quantity is insufficient, the low-risk probability is 0.1, the medium-risk probability is 0.3, and the high-risk probability is 0.6.

[0050] For each time slice, repeat the construction of the Bayesian network corresponding to that stage. The Bayesian network structures and parameters in different time slices can be appropriately adjusted according to the actual situation to reflect the dynamic changes during the construction process. A Bayesian network containing construction worker behavioral risk nodes and material property nodes is constructed within each time slice. This network links behavioral risks and material properties through causal dependency relationships, providing a basis for the probabilistic graphical model for analyzing and reasoning about risk evolution during the construction process.

[0051] S43. Add temporal dependency edges between the Bayesian networks of adjacent time slices to form the structure of the dynamic Bayesian network. Analyze the causal relationships and influence mechanisms of various risk factors and material property factors in the time dimension during the construction process to determine which factors have dependency relationships between previous and current moments. Consider the dynamic characteristics and persistence of the factors, and evaluate the influence intensity of the state at the previous moment on the state at the current moment. Based on expert knowledge and historical data, summarize significant temporal dependency relationships, such as "the behavioral risk state of personnel in the previous stage will affect the behavioral risk state of personnel in the current stage", etc.

[0052] In a dynamic Bayesian network, each time slice represents a construction stage, and there is a temporal dependence relationship between adjacent time slices. For nodes with a temporal dependence relationship, a directed edge is added between the instances at their previous and next moments, and the arrow points from the previous moment to the next moment, indicating that the state at the previous moment will affect the state at the next moment. The direction of the temporal dependence edge reflects the chronological order of the causal relationship. The node state at the previous moment is the influencing factor, and the node state at the next moment is the affected object. Each node can have multiple temporal dependence edges, which are respectively connected to different influencing factor nodes at the previous moment, indicating that the current state is comprehensively affected by multiple previous states.

[0053] The temporal dependence edge represents the influence relationship between the states at the previous and next moments, and it is necessary to quantify the intensity of this influence. The temporal dependence intensity is characterized by the conditional probability distribution parameter, that is, the probability that the node at the current moment is in various states given the node state at the previous moment. The conditional probability distribution parameter can be initially set based on expert experience and then learned and optimized using historical data (as described in step S44). A higher conditional probability value indicates that the state at the previous moment has a stronger influence on the current moment, while a lower conditional probability value indicates a weaker influence.

[0054] By adding temporal dependence edges between the Bayesian networks of adjacent time slices, a chained dynamic Bayesian network structure is formed. The dynamic Bayesian network structure consists of multiple time slices, and each time slice contains multiple nodes, representing the risk factors and material property factors within that time slice. The time slices are connected by temporal dependence edges, reflecting the dynamic evolution characteristics of risks, that is, the state at the previous stage will affect the state at the current stage.

[0055] S44. According to historical data, learn the conditional probability distribution parameters of the dynamic Bayesian network to obtain the dynamic Bayesian network model, including: organizing and annotating the historical construction data according to time slices. For each time slice, extract the behavioral risk states (low, medium, high risk) and material property states (type, quantity) therein to form a time slice - state data set. Statistically analyze the state distributions of the behavioral risk nodes and material property nodes within each time slice. For example, within a certain time slice, the proportion of low-risk behaviors is 80%, medium risk is 15%, and high risk is 5%; the quantity of steel bars is 50 tons, and the quantity of concrete is 200 cubic meters, etc. These statistical results serve as the empirical probabilities of the node states. For the temporal dependence edges between adjacent time slices, statistically analyze the joint distribution of the node states at the previous moment and the current moment. For example, the transition probability of low-risk behaviors from the previous moment to the current moment is 90% to remain low risk, 8% to turn into medium risk, and 2% to turn into high risk; the change rule of the quantity of steel bars between the previous and current moments, etc. These statistical results serve as the empirical probabilities of state transitions.

[0056] Estimate the initial state transition probability between nodes in a dynamic Bayesian network based on the empirical probabilities of node state distribution and state transition distribution, as the initial value of the conditional probability distribution parameters. Common parameter estimation methods include maximum likelihood estimation and Bayesian estimation.

[0057] Let the state of node at time t be , and its parent node at time t - 1 be . The maximum likelihood estimate of the state transition probability is: , where represents the number of times that node is observed to be in state at time t and its parent node is in state at time t - 1; represents the total number of times that the parent node is observed to be in state .

[0058] Use the EM algorithm to iteratively optimize the initial parameters to improve the accuracy of parameter estimation. At the same time, introduce a random restart strategy to avoid the parameter estimation falling into a local optimum through multiple iterations. Set the number of random restarts N, such as N = 10. For each random restart: randomly perturb the initial state transition probability to generate the initial parameter value of the EM algorithm. The perturbation can be achieved by adding a small Gaussian noise to the original estimate. E step: Use the current parameter estimate value to calculate the expected log-likelihood function of the observed data. Let the log-likelihood function of the complete data (including the observed data and the latent variables) be , then the expected log-likelihood function is: , and the expected log-likelihood function expands to: , where obs represents the observed data (state distribution and transition), and z represents the latent variables (unobserved states); represents the conditional probability of the latent variable z given the observed data obs under the current parameter ;

[0059] M step: Maximize the expected log-likelihood function to obtain the updated parameter estimate value: , repeat the EM step until the parameters converge or reach the maximum number of iterations. After all random restarts are completed, select the result with the largest expected log-likelihood function value as the optimized parameter estimate obtained by the EM algorithm.

[0060] The parameters optimized by EM are corrected using the initial state transition probability to obtain the final estimated conditional probability distribution parameters. The initial state transition probability is used as the parameter of the prior distribution . Usually, the Dirichlet distribution is chosen as the prior, and its probability density function is: , where and are the parameter vectors of the Dirichlet distribution respectively, K is the dimension of the parameter, represents the Gamma function. The parameter estimation optimized by EM is used as the likelihood of the observed data.

[0061] According to Bayes' theorem, the posterior distribution of the parameter is calculated: , , where represents the number of occurrences of the i-th parameter in the observed data.

[0062] The point estimate is extracted from the posterior distribution as the final parameter estimate value . Commonly used point estimates include the posterior mean and the posterior mode.

[0063] Posterior mean: ; Posterior mode: . Using the final conditional probability distribution parameters , a dynamic Bayesian network model is constructed. The node relationship and the temporal dependence relationship of the model have been determined in steps S42 and S43. Now, the learned numerical parameters are filled into the conditional probability distribution of each node to form a complete dynamic Bayesian network model.

[0064] S5. Use the dynamic Bayesian network risk evolution model to predict the risk distribution heat map of each construction stage in the future for a period of time, including: S51. According to the construction plan and progress, determine the future time period to be predicted, and divide the future time period into continuous time slices, where each time slice represents a construction stage. Analyze the construction plan and progress arrangement to determine the future time period for which risk prediction is required, such as the next 1 month, 3 months, or the entire construction period, etc. According to the work breakdown structure (WBS) and the critical path method (CPM) of the construction project, divide the future time period into several continuous time slices. Each time slice represents a construction stage, and the length of the time slice can be days, weeks, months, etc., which is set according to the granularity of construction management and the prediction requirements.

[0065] S52. Input the types of construction workers' behaviors and material types within each future time slice into the dynamic Bayesian network model as evidence nodes. For each future time slice, determine the types of personnel behaviors (such as normal operations, illegal operations, etc.) and material types (such as steel bars, concrete, etc.) at this stage according to the construction plan and safety management requirements. In the dynamic Bayesian network model, input these types of personnel behaviors and material types as evidence nodes. An evidence node represents known or observed information, and its state is regarded as determined during risk prediction.

[0066] S53. According to the state of the evidence nodes, use the conditional probability distribution parameters to calculate the posterior probability distribution of other risk nodes, including: taking the state of the evidence nodes as known conditions, and constructing a chain graph of the dynamic Bayesian network according to the modified conditional probability distribution parameters. According to the time slice division of the construction process, construct a chain-shaped dynamic Bayesian network diagram. Each time slice corresponds to a sub-graph, representing the static Bayesian network within that time slice. Each time slice contains multiple nodes, respectively representing construction risk factors (such as personnel behavior risks, equipment risks, environmental risks, etc.) and material attribute factors (such as material types, quantities, qualities, etc.). Within the same time slice, add directed edges between risk nodes and material attribute nodes according to the causal relationship to represent their dependence relationship. Between adjacent time slices, for each node, add a time series dependence edge pointing from the previous time slice to the next time slice to represent the influence relationship of the node between the previous and current moments. Use the modified conditional probability distribution parameters learned in step S44 to assign values to the conditional probability tables of each node to quantify the causal strength and time series dependence strength between nodes.

[0067] Select an exact algorithm suitable for dynamic Bayesian network inference, such as the Junction Tree Algorithm. This algorithm realizes efficient probability propagation and update by transforming the network into a tree structure. Instantiate the state of the evidence nodes in the end time slice as known conditions. For example, if the personnel behavior risk in a certain time slice is observed to be "high risk", then set the state of the corresponding node to "high risk". According to the instantiated state of the evidence nodes, use the conditional probability distribution parameters to calculate the posterior probability distribution of other risk nodes. The specific steps are as follows: For each non-evidence node, calculate its posterior probability in different states. The posterior probability is calculated according to Bayes' theorem: , where X is the non-evidence node, E is the evidence node, is the prior probability of X, is the conditional probability of E given X, is the marginal probability of E. Calculate using the conditional probability distribution parameters, and the prior probability is obtained according to the posterior probability or initial probability of the previous time slice. For each possible combination of states, the posterior probability is calculated to form the posterior probability distribution of non-evidence nodes. The posterior probability distributions of all non-evidence risk nodes in the terminal time slice are obtained, which characterize the probabilities of different risk factors being in various states under the known evidence.

[0068] Starting from the penultimate time slice of the chain graph, the posterior probabilities of the risk nodes in the last time slice (i.e., the terminal time slice) are used as the observed evidence. For each risk node in the penultimate time slice, the posterior probability of the corresponding node at the next moment is used as the observed value and added to the set of evidence nodes in the current time slice. Combining the original evidence node states in the current time slice and the newly added observed evidence, the inference calculation is repeated to obtain the posterior probability distributions of other risk nodes in the current time slice. Recursively advancing forward, the posterior probability distributions of the risk nodes in each time slice are calculated in turn until the starting time slice is reached.

[0069] Through recursive calculation, the risk prediction results are gradually transmitted from the back to the front, and the risk assessment of the previous moment is updated using the risk prediction information of the next moment. This recursive transmission process makes full use of the temporal dependence relationship of the dynamic Bayesian network, enabling the risk prediction to be continuously adjusted and optimized according to new observed evidence. Finally, the posterior probability distributions of the risk nodes in all time slices from the starting time slice to the terminal time slice are obtained, comprehensively characterizing the risk states and their evolution trends in each future construction stage.

[0070] S54. For each future time slice, summarize the posterior probability distributions of all risk nodes, and calculate the comprehensive probability values of the three levels of low risk, medium risk, and high risk in this time slice; arrange the comprehensive probability values of the three risk levels in each time slice in chronological order to form a probability sequence of the risk level changing with time; smooth the probability sequence of the risk level to eliminate the interference of short-term fluctuations on the long-term trend; map the smoothed probability sequence of the risk level to the time axis to generate a temporal probability distribution graph of the construction risk level; mark the risk threshold points in the temporal probability distribution graph, where the risk threshold points represent the time points when the risk level changes significantly.

[0071] S55 visualizes the time-series probability distribution diagram of the construction risk level as a risk distribution heat map; select an appropriate color scheme to map different risk levels to different colors. For example, use green, yellow, and red to represent low risk, medium risk, and high risk respectively. Convert the risk level probability of each time slice in the time-series probability distribution diagram into the corresponding color to generate a two-dimensional risk distribution heat map. The horizontal axis of the heat map represents time, and the vertical axis represents the risk level. The area size of different colors represents the probability level of the corresponding risk level. Add necessary legends, titles, and descriptions to the heat map to clearly display the distribution of future construction risks. The heat map intuitively and vividly presents the evolution trend of construction risks in the time dimension, facilitating managers to quickly identify high-risk intervals and key risk factors and take targeted risk control measures in a timely manner.

[0072] S6 conducts construction management based on the risk distribution heat map. According to the risk distribution heat map, focus on the areas with darker colors and higher risk levels, which represent the high-risk parts during the construction process. Pay attention to identifying the time periods when risks concentrate and break out or last for a long time, and these periods usually require key monitoring and control. Combine the construction plan and progress to judge the reasons for the emergence of high-risk areas and time periods, whether they are related to key processes, major changes and other factors. Use the dynamic Bayesian network model to track the causal relationship and time-series dependence relationship between risk nodes, and reveal the mechanism of risk transmission and evolution.

Claims

1. A method for managing mechanical and electrical engineering construction information, characterized in that: include: S1, collects the image set of the mechanical and electrical engineering construction site; S2, performing feature recognition on the image set, including identifying the construction worker behavior and materials, and obtaining recognition results; S3, based on the recognition results, classify the risk level of construction personnel behavior and mark material attributes; S4, divide the construction process into multiple stages, construct a Bayesian network model of time slices in each stage, add temporal dependency between the Bayesian network models of adjacent time slices, and establish a dynamic Bayesian network risk evolution model for different stages of the construction process; wherein, the construction personnel behavior risk level and material attributes are used as Bayesian network nodes, the conditional probability distribution between nodes is set, and the risk probability distribution under each factor combination is calculated by the Bayesian network inference algorithm, including: S41, according to the construction plan and progress, the entire construction process is divided into multiple continuous time slices; each time slice represents a construction stage; S42, in each time slice, the construction personnel behavior risk is divided into three levels of low risk, medium risk and high risk, as the behavior risk node of the Bayesian network; the type and quantity of materials are marked according to the material type, as the material attribute node of the Bayesian network; a Bayesian network is constructed based on the behavior risk node and the material attribute node; S43, add temporal dependency edges between the Bayesian networks of adjacent time slices to obtain a dynamic Bayesian network; wherein the temporal dependency edge represents the impact of the construction risk state at the previous moment on the current moment; S44, based on historical data, learning the conditional probability distribution parameters of the dynamic Bayesian network to obtain a dynamic Bayesian network model, including: organizing and labeling the historical data according to time slices, counting the state distribution of the behavior risk nodes and material attribute nodes in each time slice in the historical data, and obtaining the empirical probability of the node state; for the time-dependent edges between adjacent time slices, counting the joint distribution of the node state at the previous moment and the node state at the current moment, and obtaining the empirical probability of state transfer; based on the state distribution and the joint distribution, estimating the initial state transfer probability between nodes in the dynamic Bayesian network as the initial value of the conditional probability distribution parameter; using the expectation maximization EM algorithm and the random restart strategy to iteratively optimize the conditional probability distribution parameters of the dynamic Bayesian network; using the initial state transfer probability to correct the iteratively optimized conditional probability distribution parameters to obtain the corrected conditional probability distribution parameters; using the final conditional probability distribution parameters to construct a dynamic Bayesian network model; S5, using the dynamic Bayesian network risk evolution model to predict the risk distribution heat map of each stage of construction in the future, including: S51, according to the construction plan and progress, determine the future time period that needs to be predicted, and divide the future time period into continuous time slices, each time slice represents a construction stage; S52, input the construction personnel behavior type and material type in each future time slice into the dynamic Bayesian network model as an evidence node; S53, according to the state of the evidence node, use the conditional probability distribution parameters to calculate the posterior probability distribution of other risk nodes; S54, according to the posterior probability distribution of the risk node in each future time slice, generate a time series probability distribution map of the construction risk level; S55, visualize the time series probability distribution map of the construction risk level as a risk distribution heat map; S6, conduct construction management based on the risk distribution heat map.

2. The method for managing electromechanical engineering construction information according to claim 1, characterized in that: S2, perform feature recognition on the image set, including identifying the construction worker behavior and materials, and obtain recognition results, including: S21, using a convolutional neural network to extract deep features of the image as the first feature; S22, extracting texture features, edge features and color features of the image as second features; S23, fusing the first feature and the second feature to obtain a fused feature vector; S24, taking the fused feature vector as input, using the pre-trained convolutional neural network to perform forward propagation calculation, and judging the construction worker behavior type and material type in the image according to the category probability of the output layer.

3. The method for managing electromechanical engineering construction information according to claim 2, characterized in that: S3, based on the identification results, divide the construction personnel behavior risk level and mark the material attributes, including: The behavior risks are divided into three levels: low risk, medium risk and high risk according to the behavior types of construction workers. The behavior risks include not wearing safety equipment and improper operation. Mark the type and quantity of materials according to the material type.

4. The method for managing electromechanical engineering construction information according to claim 1, characterized in that: The expectation maximization EM algorithm and random restart strategy are used to iteratively optimize the conditional probability distribution parameters of the dynamic Bayesian network, including: Set the number of random restarts N. Random restarts mean reinitializing parameters during the iterative optimization process. For each random restart, the initial state transition probability is randomly perturbed to generate the initial parameter values ​​of the EM algorithm; In step E, the expected log-likelihood function of the observed data is calculated using the current conditional probability distribution parameters; the observed data includes state distribution and state transition; In the M step, the expected log-likelihood function is maximized and the estimate of the conditional probability distribution parameters is updated; Repeat steps E and M until the conditional probability distribution parameters converge or the maximum number of iterations is reached; After completing N random restarts, the result with the largest expected log-likelihood function value is selected as the conditional probability distribution parameter after iterative optimization.

5. The method for managing electromechanical engineering construction information according to claim 4, characterized in that: The initial state transition probability is used to correct the iteratively optimized conditional probability distribution parameters to obtain the corrected conditional probability distribution parameters, including: The initial state transition probability is used as the parameter of the Dirichlet distribution to construct the prior distribution of the conditional probability distribution parameters; The conditional probability distribution parameters after iterative optimization are used as observation data; According to Bayes' theorem, the posterior distribution of the conditional probability distribution parameters is calculated based on the prior distribution and the observed data; The mean or mode of the posterior distribution is used as the modified conditional probability distribution parameter.

6. The method for managing electromechanical engineering construction information according to claim 1, characterized in that: S53, according to the state of the evidence node, using the conditional probability distribution parameters, calculate the posterior probability distribution of other risk nodes, including: Taking the state of the evidence node as a known condition, a chain graph of a dynamic Bayesian network is constructed according to the modified conditional probability distribution parameters; the chain graph is composed of multiple time slices, and the nodes in each time slice represent construction risk factors and material attribute factors; From the end time slice of the chain graph, the posterior probability of other risk nodes is calculated according to the status of the evidence node and the conditional probability distribution parameters; For the previous time slice, the posterior probability of the risk node of the next time slice is used as the observation evidence, and the posterior probability of other risk nodes in the current time slice is calculated by combining the evidence node status and conditional probability distribution parameters of the current time slice; The calculation is performed recursively until the starting time slice is reached, and the posterior probability distribution of risk nodes in all time slices is obtained.

7. A computer-readable storage medium storing computer instructions, which implement the method according to any one of claims 1 to 6 when executed by a processor.

Images